摘要
问题分类是问答系统技术处理的基础与核心,它决定答案抽取的范围和方法,进而影响整个系统的性能。本文提出了一个基于贝叶斯理论的问题分类计算模型,并给出其详细算法。研究分析了问句内部结构与问题类型之间的关系,将基于疑问词的2-gram 组合和问句特征项同义近义扩展应用到具体计算中。实验表明,效果较为理想。
Question classification is the basic and core of question answering system process. It rules answer extraction range and method, and effects entire system performance. This paper proposes a new Bayes-based question classification computation modal and its detail algorithm, studies and analyzes relation of question structure and question type, and applies interrogative-based 2-gram and feature vector synonym to extend classification computation. Experiment shows that result is ideal.
出处
《计算机科学》
CSCD
北大核心
2006年第4期9-12,共4页
Computer Science
基金
基金项目:国家自然科学基金(60272088)
关键词
问答系统
问题分类
贝叶斯模型
Question answering, Question classification, Bayes model
